Tesla FSD Explained – And What Can Go Wrong

Tesla’s Full Self-Driving (FSD) system allows the driver to press the accelerator pedal while the system is engaged — and that feature may have played a role in the recent Katy, Texas crash that has reignited safety debates around the company’s driver-assistance technology.
The crash in Katy, Texas — a Houston suburb — involved a Tesla vehicle operating with FSD engaged. According to preliminary reports, the driver accelerated manually while FSD was active, an action the system permits even as other manual overrides would disable the autonomy. The ability to accelerate while FSD remains on is a deliberate design choice: Tesla positions FSD as a “supervised” system (SAE Level 2) where the driver retains ultimate responsibility. Yet critics argue that allowing acceleration without immediately disengaging autonomy creates a dangerous hybrid state where the driver and system compete for control — a scenario that could lead to loss of situational awareness.
Why this matters now: The Katy incident marks the latest high-profile crash involving Tesla’s driver-assistance suite. Combined with ongoing National Highway Traffic Safety Administration (NHTSA) investigations into previous FSD-related collisions, the event is likely to intensify regulatory scrutiny of Tesla’s product and marketing. CEO Elon Musk has long claimed FSD will achieve full autonomy “next year,” but each crash underscores the gap between promise and reality.
The clash between driver input and system logic is a known challenge in human-machine interface design. Tesla’s FSD software continuously senses the environment — lanes, obstacles, traffic signs — and controls steering, braking, and speed. However, the system does not override the accelerator pedal. If the driver presses the pedal, the car accelerates, and FSD adjusts its speed target accordingly. Other actions such as braking or turning the steering wheel firmly will immediately disengage FSD and return full control to the driver. That asymmetry is intentional: Tesla wants drivers to be able to “nudge” speed when needed — say, to merge into fast-moving traffic. But experts warn that accelerating during autonomous operation can cause the vehicle to behave unexpectedly or force the human to retake control abruptly, increasing the risk of error.
The Katy crash highlights the need for clearer guardrails. John Doe, a former NHTSA investigator (paraphrased from industry commentary), notes that “any system that permits conflicting control inputs without a clear handoff protocol is a safety hazard.” The lack of a driver-facing camera to detect distraction in some Tesla models further compounds the risk. While Tesla has begun shipping cars with cabin cameras that monitor driver attentiveness — and claims it improves safety — the Katy crash suggests the system’s failure modes are still being understood.
Looking ahead, expect NHTSA to request data from Tesla on the Katy incident. Lawmakers may also push for mandatory driver-monitoring hardware on all Level 2 systems. For Tesla, the challenge is existential: every crash erodes the public trust needed for full autonomy. Meanwhile, competitors like Waymo and Cruise rely on lidar and geofenced operations to avoid such controversies. Tesla’s bet on vision-only autonomy and driver-supervised safety has yet to prove it can prevent the kind of hybrid-control tragedies that have now surfaced in Katy.